Residual plots are widely seen in
regression
analyses. Examination of residual plot pattern can be useful in determining whether there are additional
variables
that should be included in the regression model. Residual plots also assist in outlier detection. More commonly, residual plots are used as diagnostic tools in deciding whether a
distribution
or
model
fit the data well. In linear regression,
residuals
are assumed to be normally distributed. Therefore, for convenience, they are transformed to the standardized form in
standardized residual plots
.
In the example below, the standardized residual
quantile
plot is shown in parallel with a histogram of counts. The
median
value of x corresponds to minimal fitting error in the normal quantile plot.
A plot of
bivariate
fit of the standardized residuals over a prediction interval is shown below. This plot reveals a greater fitting error at
x
=5 compared to
x
=8.